Thomas Klimpel wrote:
however i do not object to begin treating integers in semantically more correct way even if it can cause consequences
The introduction of some of the missing "standard library" functions like "modulo", "add_with_carry" or "multiply_full_result" would be a good start, and would cause nearly no harm. Of course it would only make sense if it would really enable to write fast bigint libraries in a portable way without falling back to assembler.
There is nothing stopping us from writing a boost library that does just that, implementing it as portable C++ that is slow and providing inline assmeber (or intrinsics based) implementations for common platforms. Then we can make a proposal to add these to the std library. Anyone want to suggest it as a GSOC project they are willing to mentor? I can't mentor more than one project, unfortunately. Regards, Luke